EViews.com

HI GUYS!

I am trying to run a VAR model for a group of six macroeconomic variables. To test for a co integrating vector i ran them through the JOHANSEN CO INTEGRATION TEST . COULD SOMEONE INTERPRET THE RESULTS FOR ME? should i do a VECM or a unrestricted VAR ? IT says i have 5 co-integrating vectors, is this normal? Please advice, thanks!

regards,
satya

Hi

How did u arrived the lag length for the Johansen cointegration test??

Did u made Var residuals test??

Hi.
As I understand it, Johansen Cointegration Test is used for series that are integrated of the same order. You may likely get spurrious result if you run the Test on series that are integrated of different orders. So the first step for you is to run an Augmented Dicker Fuller (ADF) unit root test on all the variables in your model. If it turns out that they are all stationary after fisrt difference (i.e., they are all integrated of order one (I(1)) )then it is safe for you to proceed with Johansen Cointegration Test. Depending on the lag lenght you choosed for the J.C test, if you come out with the same result as you did now, that is, having 5 cointegrating equations, the next appropriate step is to run a VECM with the same number of cointegrating equations you got in your J. C. test. The default is usally one but you have to change this to the same number you got in your J. C. test. Assume you got only one cointegrating equation in your J. C. test the you can run a VECM with one cointegrating equation. If however you get more than one cointegrating equation in your J. C. test, you should proceed to run the VECM with the number of cointegrating equation you got. When you do so, you are likely to get the same number of cointegrating equation you used to run your VECM in the result that comes out. The only problem you may have is how to interpret a VECM with more than one cointegrating equation. If you however find out how to do so, please share your findings with me.

Very general reply: FIRST you correctly specify your VAR model and then you use it for co-integration tests.

Very general reply: FIRST you correctly specify your VAR model and then you use it for co-integration tests.